This application proposes to use multiple modeling frameworks to examine interaction between HIV prevention interventions delivered in a combination package among a hard to reach population, people who inject drugs (PWID), in a low-and-middle income country setting (LMIC), India. Combination prevention has been recognized as the key to eradication of HIV. Critical to the determination of success of such programs will be the extent to which interventions act synergistically. Transmission models, which have been important drivers of policy for HIV prevention and treatment interventions, can help to disentangle multiple dynamic effects of combination interventions. Moreover, models can be parameterized with detailed representations of networks of interaction. However, thus far, the majority of combination prevention efforts and modeling efforts have been focused on heterosexual-driven HIV epidemics. We will leverage an NIH R01 funded cluster-randomized trial of integrated care centers to control HIV in PWID (DA032059, The NCA Study; PI: Mehta, Lucas). The intervention is a combination of the nine interventions currently recommended for PWID by the World Health Organization, including HIV counseling & testing, antiretroviral therapy, opioid substitution, needle exchange, treatment of sexually transmitted infections, treatment of tuberculosis, education and condoms, delivered in an integrated care center. A baseline sample of 14,450 was accrued using respondent-driven sampling (RDS) across 15 cities in Northeastern, Northern, and Central India. Six cites have been scaled up to receive the intervention and six will receive the control condition (all services available in distinct centers. An evaluation sample of 12,000 will be collected two years after implementation of the intervention. The parent study is underway and no additional approvals will be required from the Indian government for the proposed study. Our aims are to: 1) Explain differences in prevalence and incidence of HIV in multiple populations of PWID in India using mechanistic transmission models; and 2) Estimate the impact of multiple interventions and their interaction using mechanistic transmission models. For Aim 1, we will 1) use novel statistical inference methods to infer multiple simulated networks that are consistent with the observed data from the NCA study baseline sample accrued using RDS (n=14,450); 2) build transmission models of HIV using simulated networks of PWID estimated above; and 3) perform model validation of best fitting models using repeated leave-out one site (of 15) methods. For Aim 2, we will 1) use models developed in Aim 1 to estimate the impact of combination interventions currently being implemented in the NCA study as well as other novel interventions (e.g., PrEP); and 2) validate simulated marginal benefits of interventions. Our modeling framework can be extended to incorporate interventions not being delivered as part of the package in the NCA study (e.g., PrEP), can be extended to other hard to reach populations in India as well as PWID in other LMICs and finally as the basis for future cost-effectiveness analyses for this and other settings.